Chong Cai
Analyst · Morgan Stanley
Thank you, Eddy. This is Simon here. I'll take over from ours. There has been a lot of discussion around the AI topic recently. We have been closely monitoring and evaluating applications. Let me start with our fourth -- we see AI not as a threat to our business, but that's to enhance our capabilities. For the road freight industry in which FDA operates, the emergence of AITs can significantly lower the barriers for shippers to find available carrier capacity, reduce manual costs in the matching process and improve both matching accuracy and fulfillment rates. These changes will meaningfully improve efficiency across the entire industry. We believe this transformation will create significant opportunities for us to capture additional market shares as transactions migrate from highly fragmented offline markets, including ad hoc and relationship-based trucking networks onto our platform. In our view, AI presents more opportunities than challenges for our platform for AI models to deliver meaningful results in the highly long standardized freight matching market. They must rely on large volumes of authentic, high-frequency closed loop transaction data. This includes data such as quotes completed transactions, cancellations, fulfillment records, dispute resolutions, credit behaviors and verified logistics address database. These data sets are the results of many years of operational experience and data accumulation on our platform. Let's take pricing as example first. In long-haul freight market, competitive real-time freight rates are not publicly available. The effective transaction price for each route and time period is influenced by multiple factors, including capacity availability, backhaul demand, trucker preferences and delivery time requirements. These dynamic pricing signals can only be formed and validated within the real transaction network. On our platform, truckers must complete real name registration and facial verification before logging into our app and accessing shipment information. Negotiations between shippers and truckers are conducted through our in-app messaging tools and protected communication channels. While external AI tools, if there's any, let the basic data set to perform accurate pricing. Second, in the long-haul freight matching business, where fulfillment standards are high-end operational processes are complex, transactions involving far more than simply matching information. The capability to execute this SKU is critical. While external AI tools may help a shipper quickly obtain a price quote or even contact several truckers automatically moving our shipment from posting to final delivery requires much more than prices. Effective fulfillment depends on robust platform services and dispatch capabilities, including understanding which truckers are reliable on specific routes, their likelihood of cancellation, how trucking capacity fluctuates during different time periods and maintaining a complete operational assistance from other placement to settlement to protect the interest of both shippers and truckers. In addition, long-haul freight operations frequently involve exceptions and nonstandard situations. These may include specific vehicle requirements, trucks, equipment with gates or refrigeration units. Last mile delivery address -- last-minute delivery address changes adjustments to cargo volume, highway closure and damage disputes after delivery, handling this situation requires well-established platform rules to determine responsibilities extensive historical data to assess reliability and responsive dispatch network capable of quickly arranging alternatives when disruptions occur. These are not capabilities that a stand-alone generative AI model can deliver on its own. They are built on years of operational experience and data accumulation and are precisely where our core competitive advantage lies. In addition, our platform connects a large number of shippers and truckers and through years of operation has formed a stable transaction network and credit system. Truckers and shippers not only rely on the platform to obtain orders and capacity but also depend on the platform for credit evaluation fulfillment protection, dispute resolution and dispute resolution mechanisms. This long-term accumulation of trust and ecosystem relationships is something that a stand-alone AI agent application would find difficult to replicate. Given these structural characteristics, we believe that as AI technology continues to mature, our competitive advantages will become even more pronounced. The reason is very straightforward. The more capable AI becomes, the more it depends on real transaction data and the stronger the resulting network effects. We're actively integrating AI capabilities across multiple aspects of our platform, including matching, dispatching, pricing, risk management and customer service. As mention becomes more efficient fulfillment become more reliable and exceptional handling becomes faster and more effective. Both truckers and shippers will naturally prefer to transact on our platform. This, in turn, leads to continued data accumulation and ongoing model improvement, which further strengthens our network effects and the moat around our platform. Overall, we are very optimistic about the industry transformation and the opportunities brought by the AI era, and we are fully prepared to embrace the opportunities and challenges that come with this technological shift. For us, AI represents a capability upgrade rather than a disruption to our business model. We will leverage AI capabilities to capture the broader industry opportunities it creates making our platform more efficient and improving the user experience for both shippers and truckers. At the same time, these capabilities will further strengthen our network effects, thereby reinforcing our long-term competitive advantage in the road freight market. To address your second question on our plan for AI for 2026. Yes. As I discussed earlier on the -- on our view on AI, let me walk through the progress we made over the past quarter and our plan onwards. During the fourth quarter, our AI initiatives progress from the experimental phase to broader deployment. We're currently building an AI agent framework that covers key scenarios across our platform, including shippers, dispatch operations and customer service gradually embedding AI capabilities throughout the entire transaction workflow. Starting with the user side, our focus from shippers is simplified shipping shipment posting an automated dispatch. In the fourth quarter, we launched an AI-empowered assistant that enables shippers to submit shipping requests through a simple voice input via a floating entry point in the app. The AI can then handle the entire workflow, including freight listing, trucker screening, price negotiation and order matching significantly streamlining what previously required multiple manual steps. This capability is particularly beneficial for direct shippers as it lowers the barriers to posting shipments and improved shipping efficiency helping the platform better attract and retain SME shippers. This solution also supports vcom-based shipment posting as well as API integration delivering meaningful efficiency improvements for enterprise customers that require system integration. Our pilot results so far demonstrate the effectiveness of our AI-powered dispatch system. First, AI-driven dispatch has attracted a large number of valid trucker bids, reflecting that more accurate matching is increasing truckers' willingness to accept orders. And second, the vast majority of completed transactions are now processed entirely through automated workflows and the need for manual intervention continues to decline. Compared to traditional freight listing, AI-driven dispatch is delivering superior outcomes in both transaction efficiency and fulfillment rates. In short, the AI assistant is helping shippers reduce the time required to find truckers and helping truckers improve order pickup efficiency and enhancing overall matching quality across the platform. Internally, AI has been integrated into our customer service operations, significantly improving response times and processing efficiency while also enhancing overall service stability. Looking ahead to 2026, AI will continue to serve as a core technology foundation for improving efficiency and enhancing user experience across FTA platform. As our models continue to evolve and data advantages deepen, we expect AI to unlock additional value in areas such as matching efficiency and operational cost optimization becoming an increasingly important driver of our medium long-term growth. Thank you.